1. Field of the Invention
The present invention generally relates to a device and a method for exposing an object to a charged-particle beam, and particularly relates to an improvement in an optical system configuration for the charged-particle beam.
2. Description of the Prior Art
FIG. 1 is an illustrative drawing showing a schematic configuration of a charged-particle-beam exposure device of the prior art.
The charged-particle-beam exposure device of FIG. 1 includes an electron gun 10A, electromagnetic lenses 21 through 25, a diaphragm 12 having a rectangular aperture 12a, a mask 13 having an aperture 13a, a diaphragm 14 having a round aperture 14a, a main deflector 32, a sub-deflector 33, and deflectors 26 and 34.
Between the electron gun 10A and a wafer 11 to be exposed to a charged-particle beam, the diaphragm 12, the mask 13, and the diaphragm 14 are arranged along the optical axis AX of the charged-particle beam. The electromagnetic lenses 21 through 25 are arranged such that the charged-particle beam emitted from the electron gun 10A has a cross-sectional extension as shown by a reference numeral 15. Here, the cross-sectional extension 15 of the charged-particle beam is shown enlarged in a direction perpendicular to the optical axis AX in order to clearly show the path of the charged-particle beam.
The electromagnetic lens 21 includes an electromagnetic lens 21A and an electromagnetic lens 21B arranged along both sides of the diaphragm 12. The charged-particle beam made parallel with the optical axis AX by the electromagnetic lens 21A passes through the rectangular aperture 12a of the diaphragm 12 so that the cross section of the charged-particle beam is rectangularly shaped. The electromagnetic lens 21B converges the charged-particle beam having the rectangular cross section.
The electromagnetic lens 22 includes an electromagnetic lens 22A and an electromagnetic lens 22B arranged along both sides of the mask 13. The charged-particle beam made parallel with the optical axis AX by the electromagnetic lens 22A passes through the aperture 13a of the mask 13. By the aperture 13a, the cross section of the charged-particle beam is shaped into a pattern to be formed on the wafer 11. The electromagnetic lens 22B converges the charged-particle beam passing through the aperture 13a.
The electromagnetic lens 23 forms a cross-over image of the charged-particle beam at the round aperture 14a of the diaphragm 14. The round aperture 14a of the diaphragm 14 is used for restricting an angle of the charged-particle beam incident onto the wafer 11. Also, the diaphragm 14 is used for blocking the charged-particle beam during a non-exposure period of an exposure process. The electromagnetic lens 23 and the electromagnetic lens 24 provide an image-size reduction. The electromagnetic lens 25 is an objective lens to form an image on the wafer 11.
The main deflector 32 and the sub-deflector 33 are used for scanning the charged-particle beam on the wafer 11. The deflector 34 deflects the charged-particle beam away from the round aperture 14a of the diaphragm 14 during the non-exposure period of the exposure process to block the charged-particle beam.
In a variable-rectangle exposure method, the charged-particle beam is deflected by the deflector 26 to be displaced on the mask 13. FIG. 2 is an illustrative drawing showing a displaced charged-particle beam on the mask 13. As shown in FIG. 2, a displaced cross section 15a of the charged-particle beam has a portion 16 thereof passing through the aperture 13a of the mask 13. The portion 16 is shown as a hatched region in the figure. The portion 16 passing through the aperture 13a is reduced in size to be projected onto the wafer 11. In this manner, the cross section of the charged-particle beam is shaped into a rectangle whose shape and size vary according to a voltage applied to the deflector 26.
In a block exposure method, a plurality of blocks each having a respective aperture pattern are provided on the mask 13. One of the blocks is selected, and the charged-particle beam is directed to the selected block by the deflector 26. The charged-particle beam passes through the aperture pattern of the selected block to have a cross section accordingly shaped. Then, the shaped cross-section pattern of the charged-particle beam is reduced in size to be projected onto the wafer 11. In this manner, one shot of the charged-particle beam can create a various fine pattern on the wafer 11.
In the block exposure method, an area exposed to one shot of the charged-particle beam on the mask 13 is a rectangular block including one or more apertures, and these apertures together form the aperture pattern of the block. The cross section of the charged-particle beam is shaped in advance to correspond to this rectangular block by the rectangular aperture 12a of the diaphragm 12.
FIG. 3 is an illustrative drawing showing a configuration of the electron gun 10A with a driving circuit. The electron gun 10A of FIG. 3 includes a cathode 40A, an anode 41, a Wehnelt 42A, heaters 43 and 44, leads 45 and 46, and an insulator 471. The cathode 40A has a sharply formed tip to generate a high-voltage electric field. A radius of the curvature of the tip is about 100 .mu.m. An aperture A0 of the Wehnelt 42A has a diameter L4 of about 1.5 mm. A distance L5 between the anode 41 and the Wehnelt 42A is about 15 mm.
In order to effect a thermionic emission from the cathode 40A, a direct-current power source 50 is connected between the lead 45 and the lead 46 to heat the cathode 40A. An electron beam EBO emitted from the cathode 40A needs to be accelerated up to a predetermined energy level. In order to effect this acceleration, a plus node of a direct-current power source 51 is connected to the anode 41 having an aperture A1 and to the ground, and a minus node of the direct-current power source 51 is connected to the lead 45 via an ammeter 52, a bias resistor 53, and a bias resistor 54, and connected to the lead 46 via the ammeter 52, the bias resistor 53, and a bias resistor 55.
A voltage level of the Wehnelt 42A is controlled by a control circuit 56 such that a cathode current I1 detected by the ammeter 52 becomes a predetermined amount. The voltage level of the Wehnelt 42A is lower than that of the cathode 40A, and the cathode current I1 supplied to the cathode 40A is controlled along with the voltage level of the Wehnelt 42A. A lens effect of the Wehnelt 42A makes the electron beam EBO form a cross-over image CO.
The electron gun 10A described above has the following problems.
(1) Non-Uniform Distribution of Current Density
A current density of the cross section of the charged-particle beam emitted from the electron gun 10A has substantially a Gaussian distribution as shown at the top left of FIG. 1. Thus, the diaphragm 12 of FIG. 1 cuts off slopes of the Gaussian distribution, and the charged-particle beam passing through the rectangular aperture 12a of the diaphragm 12 has a non-uniform distribution. This non-uniformity of the current density distribution creates the following problems.
(a) Decrease in Accuracy of Resist Pattern
Assume that a pattern to be formed in a 5-.mu.m-square block for one shot of the charged-particle beam has a width of 0.1 .mu.m on the wafer 11. Because of the non-uniformity of the current density distribution, there is an error of about 0.03 .mu.m between a pattern width at a center portion of the block and a pattern width at a perimeter portion of the block. This error restricts the fineness in which the pattern can be made.
Reducing the size of the block may reduce the amount of an error. When a side length of the rectangular aperture 12a of the diaphragm 12 as well as a side length of the block on the mask 13 is shortened by a ratio of 1/1.4, for example, the number of shots of the charged-particle beam required for forming a given pattern increases twofold. Thus, reducing the size of the block results in a significant reduction of the throughput.
(b) Damage to Diaphragm 12
Because the current-density distribution of the charged-particle beam has a Gaussian profile, most of the electron beam EBO emitted from the electron gun 10A is blocked by the diaphragm 12 to be wasted. When a current amount input to the diaphragm 12 is 700 .mu.A, for example, the current amount passing through the rectangular aperture 12a of the diaphragm 12 is 20 .mu.A.
Since the electron beam EBO is accelerated by a high voltage such as 50 KV applied between the cathode 40A and the anode 41, the diaphragm 12 generates a great amount of heat. Also, because of electrons hitting the diaphragm 12, contamination is attached to an edge of the rectangular aperture 12a of the diaphragm 12 to deform an exposure pattern.
Assume that the rectangular aperture 12a is a 150-.mu.m square, a block on the mask 13 is a 300-.mu.m square, and a pattern of this block gives a reduced projection of a 5-.mu.m square on the wafer 11. If a contamination of a size of 1 .mu.m is attached to an edge of the rectangular aperture 12a, an exposure pattern on the wafer 11 has an error of 0.03 .mu.m (=1.times.5/150). If the pattern has a width of 0.1 .mu.m, the error amounts to 30% of the width. A passage of time will accumulate the contamination. Thus, the life of the diaphragm 12 is shortened by the contamination.
(c) Damage to Mask 13
The diaphragm 12 is as thin as 20 .mu.m around an aperture pattern in order to form a fine pattern, and is not provided in an effective heat releasing environment. The electron beam passing through the rectangular aperture 12a and yet having a high energy is directed to the mask 13. Since the current density distribution of the electron beam is non-uniform, the current amount of the electron beam is generally determined based on a current density near the perimeter of the beam cross section, where the current density is relatively low. This leads to an exposure of the mask 13 to an excessive amount of the electron beam. With this excessive current amount, the mask 13 is excessively heated by hitting electrons. The more the degree of non-uniformity, the smaller the usable maximum amount of the electron beam.
(d) Contamination due to Diaphragm 14
In general, a cross section of an electron beam tends to be expanded by the Coulomb interaction among the electrons. Because of the non-uniformity of the current density distribution, a center portion of the electron beam tends to have a current density which is excessively great. Therefore, an expansion of the beam cross section is relatively large when the beam has the non-uniformity.
The greater the expansion of the beam cross section, the less a current can pass through the round aperture 14a of the diaphragm 14. When a current amount passing through the rectangular aperture 12a of the diaphragm 12 is 20 .mu.A, for example, the current amount passing through the round aperture 14a of the diaphragm 14 is about 10 .mu.A. Electrons which cannot pass through the round aperture 14a cause contamination to be attached to an inside surface of the electron-beam-exposure column. Also, these electrons cause electrical charge of the electron-beam-exposure column, and this electrical charge causes a disturbance of the electrical field to vent the electron beam.
As previously described, the electron beam is deflected by the deflector 34 to be blocked by the diaphragm 14 during the non-exposure period of the exposure process. Since the beam cross section is expanded as described above, the deflection amount of the deflector 34 should be increased in order to effectively block the entire electron beam. This makes it difficult to achieve a high-speed blanking operation, and leads to a reduction of the throughput.
(2) Shortened Life of Electron Gun
Some of the positive ions generated from the diaphragm 12 or the anode 41 by the hitting of electrons are driven by the high-voltage electrical field between the anode 41 and the cathode 40A. These positive ions hit hard the tip of the cathode 40A to deform the shape of the tip. This results in a deterioration of emission characteristics of the electron beam emitted from the cathode 40A. That is, the life of the electron gun 10A is shortened.
(3) Difficulties in Adjustment of Electron Gun
Since the cathode current I1 is controlled to be constant by the control circuit 56, the voltage level of the Wehnelt 42A is varied according to a change in the tip shape and temperature of the cathode 40A. Along with a variation in the voltage level, a position of the cross-over image CO and an electron-beam-emission boundary B at the tip of the cathode 40A are changed. This leads to changes in the current density distribution on the mask 13 and in the current amount of the electron beam passing through the diaphragm 14. In order to achieve an appropriate exposure, the settings of the voltage level of the direct-current power source 51 and the cathode current I1 need to be changed accordingly. Parameters involved in the control of these settings, however, are intertwined with each other, so that the adjustment is extremely difficult.
These problems described above are present not only in the variable-rectangle method or in the block-exposure method but also in a blanking-aperture-array method, which uses in the place of the mask 13 a blanking-aperture array having a number of arranged apertures to generate a set of micro electron beams forming a desired pattern on the wafer 11.
Accordingly, there is a need for an electron gun which is easy to be adjusted, has a long life, and emits an electron beam having a relatively uniform current density distribution, and, also, a need for a charged-particle-beam exposure device and a charged-particle-beam exposure method which use this electron gun.
In the following, the problem of the damage to the mask 13 will be described in further detail.
FIG. 4A is a cross-sectional view of the mask 13 taken along a plane including the optical axis AX. The aperture 13a is formed through a thin plate of silicon by a photolithography technique. Edges of the aperture 13a can be sharpened to a fineness of an atom-size level, so that a depth of the aperture 13a in a direction of the optical axis AX can be virtually zero. When an energy of the electron beam is relatively low, a projected image of the aperture 13a on the wafer 11 is very sharp. As for dimensions, the aperture 13a is a 200-.mu.m square, and a reduction ratio is 1/100, for example. When the portion 16 of the beam cross section (FIG. 2) is a 10-.mu.m square, the projected image on the wafer 11 will be a 0.1-.mu.m square.
The lower the energy of the electron beam, the greater the exposure amount by electrons forwardly scattered in a resist layer on the wafer 11 and by electrons backwardly scattered in a silicon substrate to re-enter the resist layer. Thus, the lower the energy of the electron beam, the larger the exposure area and the less concentration of the exposure-intensity distribution on the wafer 11.
In order to sharply form a fine pattern such as having a width of 0.1 .mu.m, a high-energy electron beam such as that having an energy of 50 KV should be used. In this case, the electron beam partially penetrates through thin edges of the aperture 13a, so that a current density distribution on the wafer 11 has tapering-off slopes. FIG. 4B is an illustrative drawing showing the current density distribution on the wafer 11. The tapering off of the current density distribution on the wafer 11 leads to a reduced sharpness of the exposure pattern.
Also, there is a possibility that the mask 13 is melted by the electron beam. When this happens, the aperture 13a cannot be used any more. A reduction in the current amount of the electron beam can prevent the melting of the mask 13, but leads to a longer exposure time to assure a required exposure amount, thereby decreasing the throughput. The higher the energy of the electron beam, the smaller the exposure area on the wafer 11. Thus, a higher energy of the electron beam leads to an increased reduction in the throughput.
FIG. 4C is a cross-sectional view of an aperture formed by coating a Ta layer on the mask 13. An aperture 13b of FIG. 4C includes the mask 13 and a Ta layer 131 coated on the mask 13. Ta is a heavy metal having a high melting point, having a feature of keeping a sturdy contact with Si, and having a thermal expansion coefficient close to that of Si. Electrons of 50 KV travel 16.9 .mu.m on average in Si before they are stopped by collisions. Whereas electrons of 50 KV travel only 2.4 .mu.m on average in Ta. The melting point of Si is 1410.degree. C., while that of Ta is as high as 2990.degree. C.
When the electron beam of 50 KV is used, the electrons hit the Ta layer 131 so severely, and a slight difference in thermal expansion coefficients between the Ta layer 131 and the silicon becomes significant. As a result, the Ta layer 131 is broken off from the mask 13, so that the edges of the aperture 13a are melted by the electron beam.
Mo is known as a heavy metal having a high melting point and easily processed. When an aperture is formed through Mo, however, curvatures having a radius ranging from 10 .mu.m to 20 .mu.m are created at the corners of the aperture. These curvatures are out of a tolerance range of 0.5 .mu.m.
There is a method of obviating this problem, which is disclosed in Japanese Laid-open Patent Application No. 59-111326. FIGS. 5A through 5C are illustrative drawings showing the method of creating a rectangular aperture without the aperture rounding curvatures. As shown in FIGS. 5A and 5B, a slit 132 and a slit 133 are formed through a mask 13C1 and a mask 13C2, respectively. Then, as shown in FIG. 5C, the mask 13C1 and the mask 13C2 are overlaid one over the other to form a mask 13C such that the slits 132 and 133 form a cross. The aperture 13a of the mask 13C is a sharp rectangle without curvatures at the corners.
The slits 132 and 133 should have a width in a range between 200 .mu.m and 500 .mu.m, and, unfortunately, such slits cannot be formed by a mechanical process. Instead of a mechanical process, a dry etching process needs to be used. Thus, the thickness of the masks 13C1 and 13C2 needs to be as thin as an order of ten .mu.m. With this thickness, the mask 13C are too fragile, and it is difficult to mount the mask 13C to a holder without distorting the mask 13C.
Japanese Laid-open Patent Application No. 59-111326 also discloses a formation of a rectangular aperture by overlaying four Mo discs one over another. In this formation, edges of the rectangular aperture are too thick to form a sharp exposure image on the wafer.
Accordingly, there is a need for a charged-particle-beam exposure device and a charged-particle-beam exposure method which can use a high-energy-charged-particle beam to achieve a sufficient sharpness of an exposure pattern, and a need for a sturdy mask easy to be manufactured and a method of manufacturing the sturdy mask.
In the following, some of the above-identified problems will be described further in detail.
FIG. 6 is an illustrative drawing showing an exposure-column unit 110 of a charged-particle-beam exposure device of the block exposure type of the prior art.
In FIG. 6, the exposure-column unit 110 includes a charged-particle-beam generator 114 having a cathode 111, a grid (Wehnelt) 112, and an anode 113. The exposure-column unit 110 further includes a first slit 115 providing a rectangular shape to the charged particle beam, and a first lens 116 converging the shaped beam. The exposure-column unit 110 further includes second and third lenses 118 and 119 opposing each other, a mask 120 mounted movably in a horizontal direction between the second and third lenses 118 and 119.
On the mask 120 are provided a plurality of blocks having various aperture patterns. One of the blocks are selected, and first-to-fourth deflectors 121 through 124 deflect the beam to the selected block. The charged-particle beam passing through an aperture pattern of the selected block has a cross section shaped into the aperture pattern.
The exposure-column unit 110 further includes a blanking 125 blanking or passing the beam according to a blanking signal, a fourth lens 126 converging the beam, a round aperture 127, and a fifth lens 129. The exposure-column unit 110 further includes an objective lens 132 projecting the beam onto a wafer W, and a main deflector 133 and a sub-deflector 134 positioning the beam on the wafer W. The exposure-column unit 110 further includes a stage 135 carrying the wafer W to move it in horizontal directions.
The configuration of FIG. 6 can be used for the variable-rectangle method and the blanking-aperture array method by replacing the mask 120 with a respective mask.
In the charged-particle-beam exposure device of the prior art, elements restricting the current amount of the electron beam emitted from the electron-beam generator 114 include the first slit 115, the mask 120, and the round aperture 127.
The electron beam having a current amount of several hundreds of .mu.A when emitted from the electron-beam generator 114 is partially cut off by the first slit 115, such that the electron beam passing through the first slit 115 has a current amount of several tens of .mu.A. This electron beam with the current amount of several tens of .mu.A is directed to the mask 120. When an applied voltage is 50 KV and the current amount is .mu.A, for example, the mask 120 may generate heat of 1.0 W.
Among the three elements restricting the current amount of the electron beam, the first slit 115 and the round aperture 127 are made of metal such as molybdenum or tungsten. It is not likely that they are melted by heat. The mask 120 is made of silicon, however, because of a need to form fine apertures based on the semiconductor technology. Since the melting point of silicon is 1440.degree. C., the mask 120 may be melted through heat generated by the electron beam exposure.
Thus, as previously described, the charged-particle-beam exposure device of the prior art has a problem in that the mask can be melted due to a large current amount of the electron beam.
The electron beam is also partially cut by the round aperture 127. The round aperture 127 serves to partially cut off a cross-over image to restrict an angle of the electron beam incident onto the wafer. Here, the cross-over image is an image of the electron-beam generator 114, and the partial cutting off of the electron beam at a position where the cross-over image is formed does not affect an image of the aperture pattern of the mask 120.
The round aperture 127 is also used for completely cutting (blanking) the electron beam. When the electron beam is to be blanked, the blanking 125 deflects the electron beam such that the electron beam is shifted away from the aperture of the round aperture 127. Unfortunately, the electron beam has a Gaussian distribution in the cross section thereof as described earlier. In order to shift the electron beam completely off the aperture of the round aperture 127, the blanking 125 needs to bring about a large deflection of the electron beam. Therefore, a high voltage needs to be applied to the blanking 125, leading to a difficulty in achieving a high-speed blanking operation.
Part of the electron beam cut off by the round aperture 127 is not used for exposing the wafer, and, thus, is an excessive portion. However, this excessive portion of the electron beam hinders the high-speed blanking operation.
Therefore, the charged-particle-beam exposure device of the prior art has a problem in that the excessive portion of the electron beam hinders the high-speed blanking operation.
Furthermore, an adverse effect of the excessive portion of the electron beam can be found in accumulation of contamination. The larger the current amount of the electron beam, the more likely the contamination such as dust floating in the exposure-column unit 110 is hit by electrons to be attached to various elements of the exposure-column unit 110. Also, it is more likely that charge is built up at the contamination attached to the various elements. Such a charge is not desirable since it distorts a trajectory of the electron beam.
Accordingly, there is a need for a charged-particle-beam exposure device in which an excessive portion of the electron beam is cut off to prevent the melting of elements, to enable a high-speed blanking operation, and to reduce a possibility of accumulation and charging up of contamination.
As can be seen from the previous description, the non-uniform current density distribution and high energy of the electron beam causes various compounding problems in the charged-particle-beam exposure device of the prior art.
In summary, there is a need for an electron gun which is easily adjusted, has a long life, and emits an electron beam having a relatively uniform current density distribution, and, also, a need for a charged-particle-beam exposure device and a charged-particle-beam exposure method which use this electron gun.
Also, there is a need for a charged-particle-beam exposure device and a charged-particle-beam exposure method which can use a high-energy-charged-particle beam to achieve a sufficient sharpness of an exposure pattern, and a need for a sturdy mask easy to be manufactured and a method of manufacturing such a sturdy mask.
Further, there is a need for a charged-particle-beam exposure device in which an excessive portion of the electron beam is cut off to prevent the melting of elements, to enable a high-speed blanking operation, and to reduce a possibility of accumulation and charging up of contamination.